Reducing Petroleum Consumption from Transportation

Citation Knittel, Christopher R. Reducing Petroleum Consumption fromTransportation. Journal of Economic Perspectives 26.1 (2012):93118.

As Published http://dx.doi.org/10.1257/jep.26.1.93Publisher American Economic Association

Version Final published versionAccessed Fri Oct 11 20:36:56 EDT 2013Citable Link http://hdl.handle.net/1721.1/75296Terms of Use Article is made available in accordance with the publisher's policy

and may be subject to US copyright law. Please refer to thepublisher's site for terms of use.

Detailed Terms

The MIT Faculty has made this article openly available. Please sharehow this access benefits you. Your story matters.

Journal of Economic PerspectivesVolume 26, Number 1Winter 2012Pages 93118

TT he United States consumes more petroleum-based liquid fuel per capita he United States consumes more petroleum-based liquid fuel per capita than any other OECD high-income country30 percent more than the than any other OECD high-income country30 percent more than the second-highest country (Canada) and 40 percent more than the third-second-highest country (Canada) and 40 percent more than the third-highest (Luxembourg). The transportation sector accounts for 70 percent of U.S. highest (Luxembourg). The transportation sector accounts for 70 percent of U.S. oil consumption and 30 percent of U.S. greenhouse gas emissions. Gasoline and oil consumption and 30 percent of U.S. greenhouse gas emissions. Gasoline and diesel fuels alone account for 60 percent of oil consumption. The economic argu-diesel fuels alone account for 60 percent of oil consumption. The economic argu-ment for seeking to reduce this level of consumption of petroleum-based liquid ment for seeking to reduce this level of consumption of petroleum-based liquid fuel begins with the externalities associated with high levels of U.S. consumption of fuel begins with the externalities associated with high levels of U.S. consumption of petroleum-based fuels.petroleum-based fuels.

First, burning petroleum contributes to local pollution. The transportation First, burning petroleum contributes to local pollution. The transportation sector accounts for 67 percent of carbon monoxide emissions, 45 percent of nitrogen sector accounts for 67 percent of carbon monoxide emissions, 45 percent of nitrogen oxide (NOoxide (NOXX) emissions, and 8 percent of particulate matter emissions. These pollut-) emissions, and 8 percent of particulate matter emissions. These pollut-ants lead to health problems ranging from respiratory problems to cardiac arrest. ants lead to health problems ranging from respiratory problems to cardiac arrest. Furthermore, automobiles emit both NOFurthermore, automobiles emit both NOXX and volatile organic compounds which, and volatile organic compounds which, combined with heat and sunlight, form ground-level ozone, or smog. The papers combined with heat and sunlight, form ground-level ozone, or smog. The papers by Currie and Walker (2011) and Knittel, Miller, and Sanders (2011) both fi nd that by Currie and Walker (2011) and Knittel, Miller, and Sanders (2011) both fi nd that decreases in traffi c reduce infant mortality.decreases in traffi c reduce infant mortality.

Second, burning a gallon of gasoline causes roughly 25 pounds of carbon Second, burning a gallon of gasoline causes roughly 25 pounds of carbon dioxide to be emitted into the atmosphere, which raises the risks of destructive dioxide to be emitted into the atmosphere, which raises the risks of destructive climate change. Greenstone, Kopits, and Wolverton (2011) estimate the social cost climate change. Greenstone, Kopits, and Wolverton (2011) estimate the social cost

Reducing Petroleum Consumption from Transportation

Christopher R. Knittel is the William Barton Rogers Professor of Energy Economics, Sloan School of Management, and a Research Associate, Center for Energy and Environmental Policy Research, Massachusetts Institute of Technology, and a Research Associate, National Bureau of Economic Research, all in Cambridge, Massachusetts. His e-mail address is knittel@mit.eduknittel@mit.edu.

doi=10.1257/jep.26.1.93

Christopher R. Knittel

94 Journal of Economic Perspectives

of carbon under a variety of assumptions. They estimate a social cost of carbon as of carbon under a variety of assumptions. They estimate a social cost of carbon as high as $65 per metric ton of carbon dioxide (and gases with an effect equivalent high as $65 per metric ton of carbon dioxide (and gases with an effect equivalent to carbon dioxide) in 2010, though their values are often in the range of $21 to to carbon dioxide) in 2010, though their values are often in the range of $21 to $35 per metric ton. In Tols (2008) metastudy of the existing literature since 2001, $35 per metric ton. In Tols (2008) metastudy of the existing literature since 2001, he fi nds the median social cost of carbon ranges from $17 to $62 per metric ton of he fi nds the median social cost of carbon ranges from $17 to $62 per metric ton of COCO2 2 equivalent.equivalent.

The U.S. dependence on imported gasoline has had other costs, too, including The U.S. dependence on imported gasoline has had other costs, too, including the military expense of trying to assure stability in oil-producing regions (for the military expense of trying to assure stability in oil-producing regions (for example, ICTA 2005), and the relationship between oil price shocks and macroeco-example, ICTA 2005), and the relationship between oil price shocks and macroeco-nomic downturns. These, too, can be viewed as negative externalities. But this paper nomic downturns. These, too, can be viewed as negative externalities. But this paper neither focuses on these various externalities and social costs, nor delves into the neither focuses on these various externalities and social costs, nor delves into the literature about quantifying them. Instead, I take their existence as largely given and literature about quantifying them. Instead, I take their existence as largely given and focus on understanding the policy tools that seek to reduce gasoline consumption.focus on understanding the policy tools that seek to reduce gasoline consumption.

Of course, an obvious starting point for economists is to look at prices: although Of course, an obvious starting point for economists is to look at prices: although the price of petroleum is set in a global market, government taxes on petroleum vary the price of petroleum is set in a global market, government taxes on petroleum vary quite substantially. Table 1 lists taxes on gasoline and diesel on a per gallon basis as quite substantially. Table 1 lists taxes on gasoline and diesel on a per gallon basis as of 2010 for OECD Category I countriesessentially the worlds most developed of 2010 for OECD Category I countriesessentially the worlds most developed economies. The United States and Canada are clearly outliers, with taxes on gasoline economies. The United States and Canada are clearly outliers, with taxes on gasoline below $1 per gallon. How do these price differences affect consumption? Figure 1 below $1 per gallon. How do these price differences affect consumption? Figure 1 is suggestive. For each of these countries, it plots the per capita petroleum-based is suggestive. For each of these countries, it plots the per capita petroleum-based liquid fuel consumption versus the gasoline price in the country, with the size of the liquid fuel consumption versus the gasoline price in the country, with the size of the bubbles proportional to population. The regression line is population weighted, but bubbles proportional to population. The regression line is population weighted, but looks similar if it is not weighted. It would require quite a bit of additional argument looks similar if it is not weighted. It would require quite a bit of additional argument and delicacy to estimate a reliable elasticity of demand from these data, but for the and delicacy to estimate a reliable elasticity of demand from these data, but for the record, the slope of a fi tted log-log regression line through these data is 1.86. If record, the slope of a fi tted log-log regression line through these data is 1.86. If one were to also include the log of income as an explanatory variable in such a one were to also include the log of income as an explanatory variable in such a regression, the coeffi cient associated with the log of gasoline prices is 1.49, while regression, the coeffi cient associated with the log of gasoline prices is 1.49, while the coeffi cient associated with the log of income is 1.05.the coeffi cient associated with the log of income is 1.05.

The relative fuel use across the United States and other OECD Category I coun-The relative fuel use across the United States and other OECD Category I coun-tries is, at least in part, a by-product of differences in the types and use of light-duty tries is, at least in part, a by-product of differences in the types and use of light-duty vehicles. Schipper (2006) reports that the average gallons-per-mile of European fl eets vehicles. Schipper (2006) reports that the average gallons-per-mile of European fl eets in 2005 was below 0.034 (29.4 miles per gallon), while the average gallons-per-mile in 2005 was below 0.034 (29.4 miles per gallon), while the average gallons-per-mile of the U.S. fl eet was 0.051 (19.6 miles per gallon). Because of differences in how fuel of the U.S. fl eet was 0.051 (19.6 miles per gallon). Because of differences in how fuel effi ciency is evaluated, this fi nding probably understates the European advantage. effi ciency is evaluated, this fi nding probably understates the European advantage. Similarly, Schipper reports that per capita miles traveled in European countries is Similarly, Schipper reports that per capita miles traveled in European countries is between 35 to 45 percent of U.S. miles traveled.between 35 to 45 percent of U.S. miles traveled.

The next four sections of this paper examine the main channels through which The next four sections of this paper examine the main channels through which reductions in U.S. oil consumption might take place: 1) increased fuel economy reductions in U.S. oil consumption might take place: 1) increased fuel economy of existing vehicles, 2) increased use of non-petroleum-based, low-carbon fuels, of existing vehicles, 2) increased use of non-petroleum-based, low-carbon fuels, 3) alternatives to the internal combustion engine, and 4) reduced vehicle miles 3) alternatives to the internal combustion engine, and 4) reduced vehicle miles traveled. I then discuss how these policies for reducing petroleum consumption traveled. I then discuss how these policies for reducing petroleum consumption compare with the standard economics prescription for using a Pigouvian tax to deal compare with the standard economics prescription for using a Pigouvian tax to deal with externalities. Taking into account that energy taxes are a political hot button with externalities. Taking into account that energy taxes are a political hot button in the United States, and also considering some evidence that consumers may not in the United States, and also considering some evidence that consumers may not

Christopher R. Knittel 95

correctly value fuel economy, I offer some thoughts about the margins on which correctly value fuel economy, I offer some thoughts about the margins on which policy aimed at reducing petroleum consumption might usefully proceed.policy aimed at reducing petroleum consumption might usefully proceed.

Improved Fuel Economy

Shortly after the oil price shocks of the 1970s, the United States adopted Shortly after the oil price shocks of the 1970s, the United States adopted Corporate Average Fuel Economy (CAFE) standards, which set minimum average Corporate Average Fuel Economy (CAFE) standards, which set minimum average fuel economy thresholds for the new vehicles sold by an automaker in a given fuel economy thresholds for the new vehicles sold by an automaker in a given

Table 1Motor Fuel Taxes for OECD Category I Countries in 2010($/gallon)

Country Gasoline Diesel

United States $0.49 $0.59Canada $0.96 $0.77New Zealand $1.20 $0.00Australia $1.34 $1.34Iceland $2.28 $2.03Japan $2.59 $1.55Korea $2.64 $1.87Spain $2.66 $2.08Hungary $2.68 $2.17Austria $2.77 $2.18Luxembourg $2.90 $1.94Czech Republic $3.04 $2.59Switzerland $3.09 $3.15Slovak Republic $3.23 $2.31Sweden $3.24 $2.56Ireland $3.41 $2.82Italy $3.54 $2.65Belgium $3.58 $2.10Denmark $3.58 $2.68Portugal $3.65 $2.28France $3.80 $2.69Greece $3.82 $2.39Norway $3.87 $2.97Finland $3.93 $2.28United Kingdom $3.95 $3.95Germany $4.10 $2.95Netherlands $4.19 $2.29

Source: Taken from an Alternative Fuels and Advanced Vehicles Data Center (AFDC) worksheet: www.afdc.energy.gov/afdc/data/docs/fuel_taxes_by_country.xls. AFDCs source is the OECD/EEA database on instruments for environmental policy: http://www2.oecd.org/ecoinst/queries/index.htm.Notes: Rates as of January 1, 2010. Data for the United States and Canada include average excise taxes at the state/provincial level. VAT is not included.

96 Journal of Economic Perspectives

year. Figure 2 shows how the standard evolved. For passenger cars, the standard year. Figure 2 shows how the standard evolved. For passenger cars, the standard increased by only 0.5 miles per gallon (MPG) from 1984 to 2010; for light-duty increased by only 0.5 miles per gallon (MPG) from 1984 to 2010; for light-duty trucks, the increase was only 3.5 MPG over this same time period. From 1978 to 1991 trucks, the increase was only 3.5 MPG over this same time period. From 1978 to 1991 the standard for light trucks differentiated between two- and four-wheel drive trucks, the standard for light trucks differentiated between two- and four-wheel drive trucks, but manufacturers could also choose to meet a combined-truck standard. By world but manufacturers could also choose to meet a combined-truck standard. By world standards, these miles-per-gallon standards are not aggressive. After accounting for standards, these miles-per-gallon standards are not aggressive. After accounting for differences in the testing procedures, the World Bank estimated that the European differences in the testing procedures, the World Bank estimated that the European Union standard was roughly 17 MPG more stringent in 2010 than the U.S. standard Union standard was roughly 17 MPG more stringent in 2010 than the U.S. standard (An, Earley, and Green-Weiskel 2011).(An, Earley, and Green-Weiskel 2011).

Manufacturers who violate the CAFE standard pay a fi ne of roughly $50 per Manufacturers who violate the CAFE standard pay a fi ne of roughly $50 per mile-per-gallon per vehicle. Historically, U.S. manufacturers have complied mile-per-gallon per vehicle. Historically, U.S. manufacturers have complied with the standard. Asian manufacturers have typically exceeded the standard with the standard. Asian manufacturers have typically exceeded the standard in each year, while European manufacturers have typically violated the CAFE in each year, while European manufacturers have typically violated the CAFE standard and paid the fi nes. Trading between manufacturers was not allowed, standard and paid the fi nes. Trading between manufacturers was not allowed, so there was no possibility for certain manufacturers to accumulate credits for so there was no possibility for certain manufacturers to accumulate credits for

Figure 1Transportation Fuel Consumption per Capita versus Fuel Price

Source: Data from Worldbank.org.Notes: Size of the circle proportional to population. The line is the fi tted value from a regression of the log of consumption on the log of price.

Gasoline price ($)

2 4 6 8

0

100

200

300

400

Gas

olin

e/D

iese

l for

tran

spor

tati

on (

gallo

ns/

year

/cap

ita) United States

CanadaLuxembourg

Australia

New ZealandIceland

Sweden Greece

Czech Republic

United Kingdom

AustriaHungarySpain

GermanyFinland NorwayNetherlands

ItalyFrance

Switzerland

JapanIreland

Denamrk

PortugalBelgium

Reducing Petroleum Consumption from Transportation 97

selling a higher proportion of fuel-effi cient cars and then selling those credits to selling a higher proportion of fuel-effi cient cars and then selling those credits to other manufacturers.other manufacturers.

Other than the fact that the standards have barely budged over the last three Other than the fact that the standards have barely budged over the last three decades, two features of the original CAFE standards reduced their effect. First, decades, two features of the original CAFE standards reduced their effect. First, sport-utility vehicles were treated as light trucks, and thus could meet a lower sport-utility vehicles were treated as light trucks, and thus could meet a lower miles-per-gallon standard than cars. Perhaps not coincidentally, in 1979 light trucks miles-per-gallon standard than cars. Perhaps not coincidentally, in 1979 light trucks comprised less than 10 percent of the new vehicle fl eet, but this share rose steadily comprised less than 10 percent of the new vehicle fl eet, but this share rose steadily and peaked in 2004 at 60 percent. Second, vehicles with a gross vehicle weight and peaked in 2004 at 60 percent. Second, vehicles with a gross vehicle weight of over 8,500 pounds, which includes many large pickup trucks and sports-utility of over 8,500 pounds, which includes many large pickup trucks and sports-utility vehicles, were exempt from CAFE standards.vehicles, were exempt from CAFE standards.

Actual new vehicle fl eet fuel economy in the United States has changed little Actual new vehicle fl eet fuel economy in the United States has changed little since the early 1980s. Figure 3 plots the fuel economy of passenger vehicles (cars) since the early 1980s. Figure 3 plots the fuel economy of passenger vehicles (cars) and light duty trucks from 1979 to 2011. The fi gure shows that while the average fuel and light duty trucks from 1979 to 2011. The fi gure shows that while the average fuel economy of both cars and trucks increased over this time period, fl eet fuel economy economy of both cars and trucks increased over this time period, fl eet fuel economy fell as consumers shifted away from cars and into trucks. The fi gure also shows fell as consumers shifted away from cars and into trucks. The fi gure also shows that during the run-up in gasoline prices beginning in 2005, fl eet fuel economy that during the run-up in gasoline prices beginning in 2005, fl eet fuel economy increased. This rise appears to have subsided by 2010.increased. This rise appears to have subsided by 2010.

Although the fuel economy of new U.S. vehicles gradually declined through Although the fuel economy of new U.S. vehicles gradually declined through the late 1980s and the 1990s, there was scope for substantial improvements. In the the late 1980s and the 1990s, there was scope for substantial improvements. In the short run, when the set of offered vehicles is fi xed, car buyers could choose vehicles short run, when the set of offered vehicles is fi xed, car buyers could choose vehicles with higher fuel effi ciency. In 2011, for example, while the mean passenger car with higher fuel effi ciency. In 2011, for example, while the mean passenger car

Figure 2U.S. CAFE Standards from 1978 to 2016

Source: Data are from the National Highway Traffi c Safety Administration.

20162012

2008

Model year

20042000

19961992

19881984

1980

Stan

dard

15

20

25

30

35

40

Passenger carsLight-duty trucks

Passenger cars

Light-duty trucks

98 Journal of Economic Perspectives

available for sale was rated at 23 miles per gallon, 10 percent of passenger cars had available for sale was rated at 23 miles per gallon, 10 percent of passenger cars had a rating of 30 MPG or above. The highest rating for 2011 was the Nissan Leaf at a rating of 30 MPG or above. The highest rating for 2011 was the Nissan Leaf at 99 MPG; the Toyota Prius had a combined fuel economy rating of 50 MPG.99 MPG; the Toyota Prius had a combined fuel economy rating of 50 MPG.

In the medium run, automakers can adjust vehicle attributes by trading off In the medium run, automakers can adjust vehicle attributes by trading off weight and horsepower for increased fuel economy. In Knittel (2011), I fi nd that weight and horsepower for increased fuel economy. In Knittel (2011), I fi nd that reducing weight by 1 percent increases fuel economy by roughly 0.4 percent, while reducing weight by 1 percent increases fuel economy by roughly 0.4 percent, while reducing horsepower and torque by 1 percent increases fuel economy by roughly reducing horsepower and torque by 1 percent increases fuel economy by roughly 0.3 percent.0.3 percent.

In the long run, manufacturers can push out the frontier. In Knittel (2011), I In the long run, manufacturers can push out the frontier. In Knittel (2011), I estimate that had manufactures put all of the technological progress observed in estimate that had manufactures put all of the technological progress observed in the market from 1980 to 2006 into fuel economy, instead of putting it into attri-the market from 1980 to 2006 into fuel economy, instead of putting it into attri-butes that increased horsepower and/or weight, average fuel economy would have butes that increased horsepower and/or weight, average fuel economy would have increased by 60 percent, instead of the 11.6 percent increase actually observed. On increased by 60 percent, instead of the 11.6 percent increase actually observed. On average, a vehicle with a given weight and engine power level has a fuel economy average, a vehicle with a given weight and engine power level has a fuel economy that is 1.75 percent higher than a vehicle with the same weight and horsepower that is 1.75 percent higher than a vehicle with the same weight and horsepower level from the previous year. While the analysis in Knittel (2011) ends in 2006, using level from the previous year. While the analysis in Knittel (2011) ends in 2006, using similar data and empirical models through model year 2011, the technological fron-similar data and empirical models through model year 2011, the technological fron-tier has shifted out at an average rate of 1.97 and 1.51 percent per year from 2006 tier has shifted out at an average rate of 1.97 and 1.51 percent per year from 2006 to 2011 for passenger cars and light-duty trucks, respectively, suggesting that no to 2011 for passenger cars and light-duty trucks, respectively, suggesting that no technological barrier has yet been reached. The greater availability of hybrids and technological barrier has yet been reached. The greater availability of hybrids and plug-in hybrids also suggests that progress is likely to continue.plug-in hybrids also suggests that progress is likely to continue.

Figure 3U.S. New Vehicle Fuel Economy from 1979 to 2011

Source: Data are from the National Highway Traffi c Safety Administration.

Fuel

eco

nom

y

15

20

25

30

35

1980

Car MPGFleet MPGTruck MPG

1990 2000 2010Year

Truck MPG

Car MPG

Fleet MPG

Christopher R. Knittel 99

Gasoline prices do seem to affect choices about which cars to buy. A number Gasoline prices do seem to affect choices about which cars to buy. A number of papers have used this variation in gasoline prices to estimate the magnitude of of papers have used this variation in gasoline prices to estimate the magnitude of this response. These papers inevitably estimate a short-run response to gasoline this response. These papers inevitably estimate a short-run response to gasoline pricesin the particular sense that the choice set of vehicles is usually held fi xed. pricesin the particular sense that the choice set of vehicles is usually held fi xed. In Busse, Knittel, and Zettelmeyer (2011), my coauthors and I estimate that over In Busse, Knittel, and Zettelmeyer (2011), my coauthors and I estimate that over the period 19992008, the market share of vehicles in the bottom quartile of fuel the period 19992008, the market share of vehicles in the bottom quartile of fuel effi ciency, among those vehicles offered in a given year, falls by nearly 24 percent effi ciency, among those vehicles offered in a given year, falls by nearly 24 percent for every $1 increase in gasoline prices. In contrast, the market share of the upper for every $1 increase in gasoline prices. In contrast, the market share of the upper quartile of vehicles as ranked by fuel effi ciency increases by over 20 percent. We quartile of vehicles as ranked by fuel effi ciency increases by over 20 percent. We also show that the market share of compact cars increases by 24 percent for every also show that the market share of compact cars increases by 24 percent for every $1 increase in gasoline prices, while the market share of sport-utility vehicles falls by $1 increase in gasoline prices, while the market share of sport-utility vehicles falls by 14 percent. Klier and Linn (2010) estimate a logit demand system and focus on the 14 percent. Klier and Linn (2010) estimate a logit demand system and focus on the effects of changes in a vehicles cost per mile on demand. They fi nd that a 5 cent effects of changes in a vehicles cost per mile on demand. They fi nd that a 5 cent increase in a vehicles cost per mile, equivalent to a $1 increase in gasoline prices for increase in a vehicles cost per mile, equivalent to a $1 increase in gasoline prices for a 20 miles-per-gallon vehicle, decreases the log of its market share by between 0.5 a 20 miles-per-gallon vehicle, decreases the log of its market share by between 0.5 and 0.8, all else equal.and 0.8, all else equal.11 In the aggregate, this translates into an increase in average In the aggregate, this translates into an increase in average fuel economy of between 0.5 and 1.2 miles per gallon for every $1 dollar increase in fuel economy of between 0.5 and 1.2 miles per gallon for every $1 dollar increase in gas prices. Again, their estimates hold the set of offered vehicles fi xed. Li, Timmins, gas prices. Again, their estimates hold the set of offered vehicles fi xed. Li, Timmins, and van Haefan (2009) fi nd similar effects.and van Haefan (2009) fi nd similar effects.

A new CAFE standard in place for 2011 seeks to increase average fuel economy A new CAFE standard in place for 2011 seeks to increase average fuel economy to roughly 34.1 miles per gallon by 2016. The Environmental Protection Agency to roughly 34.1 miles per gallon by 2016. The Environmental Protection Agency and Department of Transportation are currently in the rule-making process for and Department of Transportation are currently in the rule-making process for model years 2017 and beyond, with President Obama and 13 automakers agreeing model years 2017 and beyond, with President Obama and 13 automakers agreeing to a standard of 54.5 MPG by 2025. A number of notable changes have occurred. to a standard of 54.5 MPG by 2025. A number of notable changes have occurred. First, the mileage standards are now based to some extent on the greenhouse gas First, the mileage standards are now based to some extent on the greenhouse gas emissions of the vehicle, which can deviate from fuel economy because of ancillary emissions of the vehicle, which can deviate from fuel economy because of ancillary greenhouse gas emissions associated with, for example, air conditioner refrigerant greenhouse gas emissions associated with, for example, air conditioner refrigerant leaks. Second, the new standards are footprint-based, in which each vehicle faces a leaks. Second, the new standards are footprint-based, in which each vehicle faces a standard based on the area of the footprint of its tires; larger footprints face a lower standard based on the area of the footprint of its tires; larger footprints face a lower standard. For example, the 2011 Honda Civic coupe has a footprint of 43 square standard. For example, the 2011 Honda Civic coupe has a footprint of 43 square feet, while the 2011 Ford F-150 SuperCab has a footprint of 67 square feet. In 2016, feet, while the 2011 Ford F-150 SuperCab has a footprint of 67 square feet. In 2016, these vehicles would face fuel effi ciency standards of 41.1 MPG and 24.7 MPG, these vehicles would face fuel effi ciency standards of 41.1 MPG and 24.7 MPG, respectively.respectively.22 For more details on the fuel effi ciency rules for the next few years, see For more details on the fuel effi ciency rules for the next few years, see U.S. Energy Information Administration (2007) and U.S. Environmental Protec-U.S. Energy Information Administration (2007) and U.S. Environmental Protec-tion Agency (2010).tion Agency (2010).

Is new vehicle fuel economy of 34.1 and 54.5 miles per gallon in 2016 and 2025, Is new vehicle fuel economy of 34.1 and 54.5 miles per gallon in 2016 and 2025, respectively, attainable? If we take the average rates of technological progress from respectively, attainable? If we take the average rates of technological progress from Knittel (2011) and a new vehicle fuel economy in 2010 of roughly 29 MPG, new Knittel (2011) and a new vehicle fuel economy in 2010 of roughly 29 MPG, new

1 These coeffi cients represent the change in a vehicles log of market share when the vehicles cost per mile increases. Of course, if this is driven from changes in gasoline prices, then the cost of all vehicles cost per mile will change. This explains why these effects are so large. 2 See http://www.nhtsa.gov/cars/rules/cafe/overview.htm. The sticker fuel economy is roughly 80 percent of how the vehicle is counted for the CAFE standard.

100 Journal of Economic Perspectives

vehicle fuel economy in 2016 would be roughly 32 MPG in 2016, close to the stan-vehicle fuel economy in 2016 would be roughly 32 MPG in 2016, close to the stan-dard of 34.1 MPG. Using the estimated trade-off coeffi cients, getting to 34.1 MPG dard of 34.1 MPG. Using the estimated trade-off coeffi cients, getting to 34.1 MPG would require reducing weight and engine power by less than 6 percent. Alterna-would require reducing weight and engine power by less than 6 percent. Alterna-tively, increasing the rate of technological progress to 2.75 percent per year would tively, increasing the rate of technological progress to 2.75 percent per year would achieve the mark.achieve the mark.

And, what about the standard of 54.5 miles per gallon in 2025? Taken literally, And, what about the standard of 54.5 miles per gallon in 2025? Taken literally, it it would require fundamental changes to rates of technological progress and/or require fundamental changes to rates of technological progress and/or the size and power of vehicles. The 2025 number is a bit misleading. In the law, the the size and power of vehicles. The 2025 number is a bit misleading. In the law, the 54.5 miles-per-gallon standard is based on a calculation from the Environmental 54.5 miles-per-gallon standard is based on a calculation from the Environmental Protection Agency based on carbon dioxide tailpipe emissions. It also includes Protection Agency based on carbon dioxide tailpipe emissions. It also includes credits for many technologies including plug-in hybrids, electric and hydrogen credits for many technologies including plug-in hybrids, electric and hydrogen vehicles, improved air conditioning effi ciency, and others. On an apples-to-apples vehicles, improved air conditioning effi ciency, and others. On an apples-to-apples basis, Roland (2011) cites some industry followers that claim that the actual new basis, Roland (2011) cites some industry followers that claim that the actual new fl eet fuel economy standard in 2025 is more like 40 miles per gallon. Achieving fl eet fuel economy standard in 2025 is more like 40 miles per gallon. Achieving 40 miles per gallon by 2025 is certainly possible. At a rate of technological prog-40 miles per gallon by 2025 is certainly possible. At a rate of technological prog-ress of 1.75 percent per year, 40 miles per gallon requires additional reductions in ress of 1.75 percent per year, 40 miles per gallon requires additional reductions in weight and engine power of less than 7 percent.weight and engine power of less than 7 percent.

Alternative Fuels

Biofuels are derived from biological components like corn, soybeans, sugar, Biofuels are derived from biological components like corn, soybeans, sugar, grasses, and wood chips. The ethanol produced by this process is an imperfect grasses, and wood chips. The ethanol produced by this process is an imperfect substitute for gasoline, although biodiesel is a nearly perfect substitute for petro-substitute for gasoline, although biodiesel is a nearly perfect substitute for petro-leum-based diesel. (Methanol is another alcohol and imperfect substitute for leum-based diesel. (Methanol is another alcohol and imperfect substitute for gasoline that can be derived from either methanethat is, natural gasbiomass, or gasoline that can be derived from either methanethat is, natural gasbiomass, or coal.) Biofuels also hold the potential to have lower carbon emissions. If the plant coal.) Biofuels also hold the potential to have lower carbon emissions. If the plant material could be grown and converted to liquid fuel using only technologies that material could be grown and converted to liquid fuel using only technologies that do not produce any greenhouse gas emissions, and not lead to land use changes do not produce any greenhouse gas emissions, and not lead to land use changes that increase greenhouse gases, then biofuels would not emit any net greenhouse that increase greenhouse gases, then biofuels would not emit any net greenhouse gases over the lifecycle.gases over the lifecycle.

In practice, the lifecycle emissions of biofuels are affected by a number of In practice, the lifecycle emissions of biofuels are affected by a number of factors. First, the feedstock used affects carbon emissions during the growing factors. First, the feedstock used affects carbon emissions during the growing stagefor example, through fertilization. The most common feedstock used in stagefor example, through fertilization. The most common feedstock used in the United States is corn. Brazilian ethanol is made from sugar cane. So-called the United States is corn. Brazilian ethanol is made from sugar cane. So-called second generation or cellulosic ethanol uses feedstocks that require little in second generation or cellulosic ethanol uses feedstocks that require little in the way of irrigation and fertilizer during the growing process, such as miscanthus the way of irrigation and fertilizer during the growing process, such as miscanthus and switchgrass. Second, the fuel used for generation of heat and electricity during and switchgrass. Second, the fuel used for generation of heat and electricity during the refi ning process affects emissions. Third, the calculation is affected by whether the refi ning process affects emissions. Third, the calculation is affected by whether the coproducts from distilling, notably distillery grains with solubles, are dried the coproducts from distilling, notably distillery grains with solubles, are dried before being sold and whether the emissions from drying should be included, or before being sold and whether the emissions from drying should be included, or treated as another product.treated as another product.

Fourth, the lifecycle emissions of corn-based biofuels are affected by the milling Fourth, the lifecycle emissions of corn-based biofuels are affected by the milling process. Corn ethanol is typically refi ned using either a dry or wet milling process. process. Corn ethanol is typically refi ned using either a dry or wet milling process.

Reducing Petroleum Consumption from Transportation 101

Under wet milling, the corn is soaked in hot water and sulfurous acid. The starches Under wet milling, the corn is soaked in hot water and sulfurous acid. The starches from this mixture are then separated and fermented, leading to ethanol. Dry milling from this mixture are then separated and fermented, leading to ethanol. Dry milling requires less energy and generates fewer greenhouse gas emissions, but does not requires less energy and generates fewer greenhouse gas emissions, but does not yield as many coproducts as wet milling. Under dry milling, the corn is ground into yield as many coproducts as wet milling. Under dry milling, the corn is ground into fl our and cooked along with enzymes, where yeast is added for fermentation. The fl our and cooked along with enzymes, where yeast is added for fermentation. The ethanol is then separated from the liquid. The remaining component undergoes ethanol is then separated from the liquid. The remaining component undergoes another process turning it into livestock feed.another process turning it into livestock feed.

Finally, and most diffi cult to estimate, increases in biofuel production can alter Finally, and most diffi cult to estimate, increases in biofuel production can alter land use patterns elsewhere. For example, an increase in Brazilian sugar cane ethanol land use patterns elsewhere. For example, an increase in Brazilian sugar cane ethanol may reduce pasturelands and thus cause cattle farmers to cut down rainforest, may reduce pasturelands and thus cause cattle farmers to cut down rainforest, which reduces the quantity of greenhouse gases sequestered by the rainforest. The which reduces the quantity of greenhouse gases sequestered by the rainforest. The infl uential paper by Searchinger et al. (2008) was the fi rst to measure this factor, infl uential paper by Searchinger et al. (2008) was the fi rst to measure this factor, fi nding that once indirect land use effects are considered, corn-based ethanol can fi nding that once indirect land use effects are considered, corn-based ethanol can have nearly twice the greenhouse gas emissions of gasoline. A number of follow-up have nearly twice the greenhouse gas emissions of gasoline. A number of follow-up papers have found that while these effects may not be this large, they remain impor-papers have found that while these effects may not be this large, they remain impor-tant. For example, Tyner, Taheripour, Zhuang, Birur, and Baldos (2010) argue that tant. For example, Tyner, Taheripour, Zhuang, Birur, and Baldos (2010) argue that once changes in both international trade and crop yields are accounted for, corn once changes in both international trade and crop yields are accounted for, corn ethanol results in fewer greenhouse gas emissions than gasoline, despite indirect ethanol results in fewer greenhouse gas emissions than gasoline, despite indirect land use changes.land use changes.

How does the sum of these factors compare to the emissions of gasoline? The How does the sum of these factors compare to the emissions of gasoline? The emissions of a gallon of gasoline over the entire lifecycle of its production depend emissions of a gallon of gasoline over the entire lifecycle of its production depend on, amongst other things, the effi ciency of the refi nery and weight of the oil. A on, amongst other things, the effi ciency of the refi nery and weight of the oil. A number of estimates exist. The California Air Resource Board (2011) estimated number of estimates exist. The California Air Resource Board (2011) estimated that an average gallon of California-refi ned gasoline generates 27.9 pounds of that an average gallon of California-refi ned gasoline generates 27.9 pounds of COCO22 -equivalent greenhouse gas emissions. Roughly 19 pounds of this comes from -equivalent greenhouse gas emissions. Roughly 19 pounds of this comes from the combustion of the gasoline, while the remainder comes from the emissions the combustion of the gasoline, while the remainder comes from the emissions associated with refi ning, transporting, and so on. The 19 pounds fi gure may sound associated with refi ning, transporting, and so on. The 19 pounds fi gure may sound too high, given that a gallon of gasoline weighs roughly 6 pounds. The reason is too high, given that a gallon of gasoline weighs roughly 6 pounds. The reason is that during the combustion process the carbon atoms in the gasoline, which have a that during the combustion process the carbon atoms in the gasoline, which have a molecular weight of 12, combine with 2 oxygen atoms from the atmosphere, each molecular weight of 12, combine with 2 oxygen atoms from the atmosphere, each having a molecular weight of 16.having a molecular weight of 16.

The California Air Resource Board (2011) also estimates that lifecycle emissions The California Air Resource Board (2011) also estimates that lifecycle emissions for a number of ethanol pathways lead to for a number of ethanol pathways lead to higher greenhouse gas emissions than gaso-r greenhouse gas emissions than gaso-line. For example, Midwest ethanol (shipped to California) produced using a wet line. For example, Midwest ethanol (shipped to California) produced using a wet mill process and coal for heating and electricity has 26 percent more greenhouse gas mill process and coal for heating and electricity has 26 percent more greenhouse gas emissions than the average gasoline refi ned in California. In contrast, dry mill, wet emissions than the average gasoline refi ned in California. In contrast, dry mill, wet distillery grains with solubles Californian ethanol which uses 80 percent natural gas distillery grains with solubles Californian ethanol which uses 80 percent natural gas and 20 percent biomass is predicted to have greenhouse emissions that are 19 percent and 20 percent biomass is predicted to have greenhouse emissions that are 19 percent below that of gasoline. Brazilian ethanol made from sugarcane has the lowest lifecycle below that of gasoline. Brazilian ethanol made from sugarcane has the lowest lifecycle emissions among those pathways analyzed in the California report. An Environmental emissions among those pathways analyzed in the California report. An Environmental Protection Agency (2009) report reaches similar conclusions. Dry mill ethanol made Protection Agency (2009) report reaches similar conclusions. Dry mill ethanol made using coal has either 13 or 34 percent using coal has either 13 or 34 percent more emissions than gasoline. However, dry mill emissions than gasoline. However, dry mill ethanol using biomass, a form of cellulosic ethanol, in a combined heat and power ethanol using biomass, a form of cellulosic ethanol, in a combined heat and power system has 26 or 47 percent fewer emissions.system has 26 or 47 percent fewer emissions.

102 Journal of Economic Perspectives

In short, lifecycle analyses suggest that corn-based ethanol can play only a In short, lifecycle analyses suggest that corn-based ethanol can play only a marginal role in reducing greenhouse gas emissions from the transportation marginal role in reducing greenhouse gas emissions from the transportation sector. In contrast, cellulosic-based biofuels can potentially play a much larger role, sector. In contrast, cellulosic-based biofuels can potentially play a much larger role, although there remain technological obstacles to widespread mass production of although there remain technological obstacles to widespread mass production of ethanol at low cost from this source.ethanol at low cost from this source.

There are other natural limits to the impact of corn-based ethanol production There are other natural limits to the impact of corn-based ethanol production in the United States as well. How much farmland would be required if Americas in the United States as well. How much farmland would be required if Americas cars were to run solely on E85, which is 85 percent ethanol and 15 percent gasoline? cars were to run solely on E85, which is 85 percent ethanol and 15 percent gasoline? Well, gasoline usage in the United States is roughly 140 billion gallons per year, and Well, gasoline usage in the United States is roughly 140 billion gallons per year, and it takes 128,500 acres of corn to produce 50 million gallons of ethanol (according it takes 128,500 acres of corn to produce 50 million gallons of ethanol (according to the FAQ at to the FAQ at http://ethanol.orghttp://ethanol.org). Given that ethanol has an energy content that ). Given that ethanol has an energy content that is roughly 67 percent of gasoline, 140 billion gallons of our current fuel, which is is roughly 67 percent of gasoline, 140 billion gallons of our current fuel, which is roughly 5 percent ethanol, would equal roughly 190 billion gallons of E85. Thus, roughly 5 percent ethanol, would equal roughly 190 billion gallons of E85. Thus, if the ethanol used corn as the feedstock, this would imply roughly 415 million if the ethanol used corn as the feedstock, this would imply roughly 415 million acres of corn cropbut there is currently only 406 million acres of farmed land in acres of corn cropbut there is currently only 406 million acres of farmed land in the United States. In short, signifi cant expansion of corn-based ethanol production the United States. In short, signifi cant expansion of corn-based ethanol production is likely to require additional land, which unleashes environmental consequences is likely to require additional land, which unleashes environmental consequences discussed earlier. In addition, corn-based biofuels also compete with current uses of discussed earlier. In addition, corn-based biofuels also compete with current uses of corn, which has implications for the worldwide price of corn and other substitute corn, which has implications for the worldwide price of corn and other substitute grains. Cellulosic biofuels, in contrast, offer a feedstock that will not compete with grains. Cellulosic biofuels, in contrast, offer a feedstock that will not compete with food products nearly as much, since these plants can be grown on marginal lands food products nearly as much, since these plants can be grown on marginal lands and without irrigation.and without irrigation.

Large-scale substitution of ethanol for gasoline is limited in the short run Large-scale substitution of ethanol for gasoline is limited in the short run because of the blend wallthe percentage of fuel that can be ethanol and safely because of the blend wallthe percentage of fuel that can be ethanol and safely burned in a vehicle designed to burn only gasoline. The Environmental Protection burned in a vehicle designed to burn only gasoline. The Environmental Protection Agency recently ruled that vehicles of model year 2005, or newer, can safely burn Agency recently ruled that vehicles of model year 2005, or newer, can safely burn fuel that is 15 percent ethanol. Vehicles older than this can burn E10. Flex-fuel fuel that is 15 percent ethanol. Vehicles older than this can burn E10. Flex-fuel vehicles, in contrast, can burn fuel that is up to 85 percent ethanol.vehicles, in contrast, can burn fuel that is up to 85 percent ethanol.

U.S. policymakers have adopted a variety of biofuel policies: performance stan-U.S. policymakers have adopted a variety of biofuel policies: performance stan-dards, subsidies, and mandates. The Volumetric Ethanol Excise Tax Credit expired dards, subsidies, and mandates. The Volumetric Ethanol Excise Tax Credit expired on December 31, 2011. The credit offered fuel blenders $0.45 tax credit per gallon on December 31, 2011. The credit offered fuel blenders $0.45 tax credit per gallon of ethanol sold. Before this tax credit, ethanol received an implicit subsidy (relative of ethanol sold. Before this tax credit, ethanol received an implicit subsidy (relative to gasoline) as it was exempted from the federal fuel excise tax in 1978. The 2008 to gasoline) as it was exempted from the federal fuel excise tax in 1978. The 2008 Farm Bill differentiated between corn-based and cellulosic ethanol, with cellulosic Farm Bill differentiated between corn-based and cellulosic ethanol, with cellulosic ethanol receiving a $0.91 per gallon tax credit, minus an applicable tax credit ethanol receiving a $0.91 per gallon tax credit, minus an applicable tax credit collected by the blender of the cellulosic ethanol.collected by the blender of the cellulosic ethanol.33 Small ethanol producersthose Small ethanol producersthose with a capacity of less than 60 million gallonsreceived an additional 10 cents per with a capacity of less than 60 million gallonsreceived an additional 10 cents per gallon credit.gallon credit.

3 These fi gures understate the subsidy level because they are on a per-gallon basis, not on a per-energy basis. As noted in the text, one gallon of ethanol has roughly 67 percent of the energy content of a gallon of gasoline, implying that it requires 1.48 gallons of ethanol to displace one gallon of gasoline. Therefore, on a per gallon of gasoline equivalent basis, corn-based ethanol received a 67 cents per gallon of gaso-line equivalent subsidy; 81 cents for a small producer. Cellulosic ethanol received a $1.35 per gallon of gasoline equivalent subsidy, $1.49 per gallon of gasoline equivalent for small producers.

Christopher R. Knittel 103

Similar subsidies existed for biodiesel. The Jobs Creation Act of 2004 established Similar subsidies existed for biodiesel. The Jobs Creation Act of 2004 established a $1-per-gallon tax credit for biodiesel created from virgin oil, defi ned as oil coming a $1-per-gallon tax credit for biodiesel created from virgin oil, defi ned as oil coming from animal fats or oilseed rather than recycled from cooking oil. Biodiesel from from animal fats or oilseed rather than recycled from cooking oil. Biodiesel from recycled oil receives a $0.50-per-gallon tax credit. These subsidies were extended recycled oil receives a $0.50-per-gallon tax credit. These subsidies were extended under the Energy Policy Act of 2007, but also expired at the end of 2011.under the Energy Policy Act of 2007, but also expired at the end of 2011.

The other major federal ethanol policy is mandates to use such fuels. The The other major federal ethanol policy is mandates to use such fuels. The fi rst Renewable Fuel Standard was adopted in 2005. The Energy Independence fi rst Renewable Fuel Standard was adopted in 2005. The Energy Independence and Security Act of 2007 expanded this standard by calling for 36 billion gallons and Security Act of 2007 expanded this standard by calling for 36 billion gallons of biofuelsincluding 21 billion gallons of advanced biofuels by 2022, which of biofuelsincluding 21 billion gallons of advanced biofuels by 2022, which are to have a lower greenhouse gas content than corn-based ethanol. Given how are to have a lower greenhouse gas content than corn-based ethanol. Given how the Renewable Fuel Standard is implemented, ethanol prices refl ect an implicit the Renewable Fuel Standard is implemented, ethanol prices refl ect an implicit subsidy, while gasoline is priced as if it were taxed (Holland, Hughes, Knittel, and subsidy, while gasoline is priced as if it were taxed (Holland, Hughes, Knittel, and Parker 2011). A variety of state-level blend minimums and performance standards Parker 2011). A variety of state-level blend minimums and performance standards also exist.also exist.

Methanol is another alcohol that can be used as a liquid fuel. Methanol Methanol is another alcohol that can be used as a liquid fuel. Methanol production is an established industry: methanol is used as a racing fuel, as an indus-production is an established industry: methanol is used as a racing fuel, as an indus-trial chemical, and as a liquid fuel in some countriesespecially China. Methanol trial chemical, and as a liquid fuel in some countriesespecially China. Methanol can be produced from natural gas, coal, or biomass. In 2010, the United States can be produced from natural gas, coal, or biomass. In 2010, the United States consumed 1.8 billion gallons of methanol with world production totaling over consumed 1.8 billion gallons of methanol with world production totaling over 15 billion gallons (see statistics at 15 billion gallons (see statistics at http://methanol.orghttp://methanol.org), roughly on par with ), roughly on par with global ethanol production of 23 billion gallons in 2010 (see statistics at global ethanol production of 23 billion gallons in 2010 (see statistics at http://http://ethanolproducer.comethanolproducer.com). In contrast to ethanol, most methanol consumption is not ). In contrast to ethanol, most methanol consumption is not as a fuel, but as a chemical feedstock.as a fuel, but as a chemical feedstock.

Methanol has three main advantages over corn-based ethanol. First, on a Methanol has three main advantages over corn-based ethanol. First, on a greenhouse gas basis, Delucchi (2005) estimates that methanol produced from greenhouse gas basis, Delucchi (2005) estimates that methanol produced from natural gas has 11 percent lower greenhouse gas emissions than corn-based natural gas has 11 percent lower greenhouse gas emissions than corn-based ethanol. However, he fi nds that methanol still has higher emissions than gasoline. ethanol. However, he fi nds that methanol still has higher emissions than gasoline. Others fi nd that the greenhouse gas emissions from methanol are roughly equiva-Others fi nd that the greenhouse gas emissions from methanol are roughly equiva-lent to gasoline (MIT 2011). Second, methanol is cheaper than gasoline, at least at lent to gasoline (MIT 2011). Second, methanol is cheaper than gasoline, at least at current oil and gas prices. Methanex, the worlds largest methanol producer, quotes current oil and gas prices. Methanex, the worlds largest methanol producer, quotes current retail methanol prices in North America of $1.38 per gallon. Methanol has current retail methanol prices in North America of $1.38 per gallon. Methanol has an even lower energy content than ethanol at roughly 53 percent of gasoline, so an even lower energy content than ethanol at roughly 53 percent of gasoline, so this implies a cost per gallon of gasoline equivalent of $2.51 per gallon (approxi-this implies a cost per gallon of gasoline equivalent of $2.51 per gallon (approxi-mately, because of changes in engine effi ciency), still cheaper than gasoline. Third, mately, because of changes in engine effi ciency), still cheaper than gasoline. Third, methanol doesnt rely on crops, eliminating the negative consequences associated methanol doesnt rely on crops, eliminating the negative consequences associated with crop production.with crop production.

Methanol also faces four disadvantages. First, methanol produced from natural Methanol also faces four disadvantages. First, methanol produced from natural gas cannot achieve the same reductions in greenhouse gas emissions as second-gas cannot achieve the same reductions in greenhouse gas emissions as second-generation or cellulosic ethanol. Second, alcohols are generally more corrosive generation or cellulosic ethanol. Second, alcohols are generally more corrosive than gasoline, and methanol is even more corrosive than ethanol. For vehicles to than gasoline, and methanol is even more corrosive than ethanol. For vehicles to run on ethanol or methanol, manufacturers must protect certain engine parts and run on ethanol or methanol, manufacturers must protect certain engine parts and rubber material from the fuels. Flex-fuel vehicles that can run on fuel that is as rubber material from the fuels. Flex-fuel vehicles that can run on fuel that is as much as 85 percent methanol (M85) require a slightly larger investment, on the much as 85 percent methanol (M85) require a slightly larger investment, on the order of $200 per vehicle (MIT 2011). Third, as discussed above, methanol has an order of $200 per vehicle (MIT 2011). Third, as discussed above, methanol has an

104 Journal of Economic Perspectives

even lower energy content than ethanol, so a tank of gas wouldnt take you as far.even lower energy content than ethanol, so a tank of gas wouldnt take you as far.44 Finally, there are open questions as to how safe the drilling process is, or can be, for Finally, there are open questions as to how safe the drilling process is, or can be, for shale gas, including potential problems of methane leakage.shale gas, including potential problems of methane leakage.

Given the recent discoveries of large shale gas deposits within North America, Given the recent discoveries of large shale gas deposits within North America, a compelling argument can be made that methanol, as a substitute for gasoline, a compelling argument can be made that methanol, as a substitute for gasoline, should have the same support as corn-based ethanol. Methanol carries similar should have the same support as corn-based ethanol. Methanol carries similar greenhouse gas reductions, if not larger, and is not petroleum based. The open greenhouse gas reductions, if not larger, and is not petroleum based. The open issue is whether drilling for shale gas has fewer environmental repercussions than issue is whether drilling for shale gas has fewer environmental repercussions than the land use implications of ethanol.the land use implications of ethanol.

An alternative use for natural gas is in compressed natural gas (CNG) vehicles, An alternative use for natural gas is in compressed natural gas (CNG) vehicles, which use internal combustion engines to burn natural gas stored at high pressures. which use internal combustion engines to burn natural gas stored at high pressures. Rood Werpy, Santine, Burnham, and Mintz (2010) summarize tailpipe emission Rood Werpy, Santine, Burnham, and Mintz (2010) summarize tailpipe emission comparisons of vehicles and fi nd that compressed natural gas has emission reduc-comparisons of vehicles and fi nd that compressed natural gas has emission reduc-tions that are often above 20 percent, compared to gasoline, but often less than tions that are often above 20 percent, compared to gasoline, but often less than 10 percent when compared to diesel fuel. Moreover, long-run average costs on a 10 percent when compared to diesel fuel. Moreover, long-run average costs on a gallon-of-gasoline equivalent are currently below gasoline: the U.S. Department of gallon-of-gasoline equivalent are currently below gasoline: the U.S. Department of Energy reports national average prices of $2.09 for October 2011.Energy reports national average prices of $2.09 for October 2011.

The drawbacks to CNG vehicles are similar to electric vehicles (discussed The drawbacks to CNG vehicles are similar to electric vehicles (discussed below). New infrastructure is needed for refueling with compressed natural gas. below). New infrastructure is needed for refueling with compressed natural gas. Refueling can take longer, especially if done at home: slow-fi ll home units can take Refueling can take longer, especially if done at home: slow-fi ll home units can take over four hours. CNG vehicles have limited range, often the equivalent of about over four hours. CNG vehicles have limited range, often the equivalent of about eight gallons of gasoline. CNG vehicles also have a higher upfront cost: the Honda eight gallons of gasoline. CNG vehicles also have a higher upfront cost: the Honda Civic GX, a CNG vehicle, sells for, roughly, a $4,000 premium but has 27 percent Civic GX, a CNG vehicle, sells for, roughly, a $4,000 premium but has 27 percent less horsepower than a comparable gasoline-powered car. A thorough comparison less horsepower than a comparable gasoline-powered car. A thorough comparison of CNG and electric vehicles is beyond the scope of this paper, but again, given of CNG and electric vehicles is beyond the scope of this paper, but again, given large natural gas reserves recently discovered, this would appear to be a worthwhile large natural gas reserves recently discovered, this would appear to be a worthwhile avenue for research. My read of the literature suggests that these drawbacks are avenue for research. My read of the literature suggests that these drawbacks are not as severe with CNG vehicles as with electric vehicles, although the reduction not as severe with CNG vehicles as with electric vehicles, although the reduction in carbon emissions from CNG vehicles may also be less than if the electricity for a in carbon emissions from CNG vehicles may also be less than if the electricity for a vehicle is generated in a low-carbon manner. Once the benefi ts of both greenhouse vehicle is generated in a low-carbon manner. Once the benefi ts of both greenhouse gas emissions and petroleum reductions are compared with the added costs, CNG gas emissions and petroleum reductions are compared with the added costs, CNG vehicles might make more sense than electric vehicles.vehicles might make more sense than electric vehicles.

Replacing the Internal Combustion Engine

Shifting away from the internal combustion engine to powering vehicles with Shifting away from the internal combustion engine to powering vehicles with electricity or with hydrogen is another way of reducing petroleum usage. Either electricity or with hydrogen is another way of reducing petroleum usage. Either approach could represent a reduction in the pollutants per unit of energy of the approach could represent a reduction in the pollutants per unit of energy of the fuel and an increase in fuel economyas measured by the energy required to travel fuel and an increase in fuel economyas measured by the energy required to travel

4 Because of their lower vapor pressure, starting an engine in cold weather is more diffi cult when using ethanol and methanol (with ethanol having a lower vapor pressure compared to methanol), which may prompt consumers to use a lower blend of these fuels during the winter.

Reducing Petroleum Consumption from Transportation 105

one mile. In terms of greenhouse gas emissions, all-electric or hydrogen vehicles one mile. In terms of greenhouse gas emissions, all-electric or hydrogen vehicles have been viewed by some as the end game, since it is possible to generate either have been viewed by some as the end game, since it is possible to generate either electricity or hydrogen in a carbon-free waysay, through solar or wind power. electricity or hydrogen in a carbon-free waysay, through solar or wind power. Ultimately, both technologies would probably use electric motors. It is possible Ultimately, both technologies would probably use electric motors. It is possible to burn hydrogen directly in an internal combustion engine. BMW, for example, to burn hydrogen directly in an internal combustion engine. BMW, for example, has a fl ex-fuel 7-series that can use both diesel and hydrogen. However, this has a fl ex-fuel 7-series that can use both diesel and hydrogen. However, this forgoes the effi ciency gain from electric motors, so most industry followers believe forgoes the effi ciency gain from electric motors, so most industry followers believe that if hydrogen were to penetrate the market it would do so through a fuel cell that if hydrogen were to penetrate the market it would do so through a fuel cell that powered an electric motor vehicle.that powered an electric motor vehicle.55

The hurdle for both electricity and hydrogen technologies is, of course, cost. The hurdle for both electricity and hydrogen technologies is, of course, cost. These costs can usefully be divided up into costs for vehicles, cost of the electricity These costs can usefully be divided up into costs for vehicles, cost of the electricity or hydrogen itself, and infrastructure costs associated with reenergizing vehicles.or hydrogen itself, and infrastructure costs associated with reenergizing vehicles.

For a pure electric vehicle, battery technology still imposes some daunting For a pure electric vehicle, battery technology still imposes some daunting constraints. While I am not aware of any studies detailing the required battery size as constraints. While I am not aware of any studies detailing the required battery size as a function of key variables such as the vehicle weight and desired range, some rough a function of key variables such as the vehicle weight and desired range, some rough calculations are possible. My personal communications with Yet-Ming Chiang of calculations are possible. My personal communications with Yet-Ming Chiang of MIT suggest that a current mid-sized sedan, weighing about 3,000 pounds, requires MIT suggest that a current mid-sized sedan, weighing about 3,000 pounds, requires roughly 300 watt-hours of battery capacity for every mile of range. This fi gure for roughly 300 watt-hours of battery capacity for every mile of range. This fi gure for a mid-sized sedan is roughly comparable to the 2011 Nissan Leaf, which weighs a mid-sized sedan is roughly comparable to the 2011 Nissan Leaf, which weighs 3,354 pounds. The Leaf has a 24-kilowatt-hour battery pack and has a range rating 3,354 pounds. The Leaf has a 24-kilowatt-hour battery pack and has a range rating of 73 miles from the Environmental Protection Agency, which translating to 328 watt of 73 miles from the Environmental Protection Agency, which translating to 328 watt hours per mile. A mid-sized sport-utility vehicle, weighing roughly 4,000 pounds, hours per mile. A mid-sized sport-utility vehicle, weighing roughly 4,000 pounds, requires 425 watt hours for every mile of range. For a 200-mile range, which is requires 425 watt hours for every mile of range. For a 200-mile range, which is signifi cantly lower than current internal-combustion-based vehicles, the mid-sized signifi cantly lower than current internal-combustion-based vehicles, the mid-sized sedan would require a 60-kilowatt-hour battery pack, while the mid-sized sport-utility sedan would require a 60-kilowatt-hour battery pack, while the mid-sized sport-utility vehicle would require a 85-kilowatt-hour battery pack. As a third point of reference, vehicle would require a 85-kilowatt-hour battery pack. As a third point of reference, a 2011 Ford F-150 SuperCab weighs 5,500 pounds. If the relationship is roughly a 2011 Ford F-150 SuperCab weighs 5,500 pounds. If the relationship is roughly linear, a pickup truck of this size would require a 123-kilowatt-hour battery pack. I linear, a pickup truck of this size would require a 123-kilowatt-hour battery pack. I should note that I am ignoring the effects of the batterys weight, which have real should note that I am ignoring the effects of the batterys weight, which have real consequences (Kromer and Heywood 2007). For example, the battery and control consequences (Kromer and Heywood 2007). For example, the battery and control module for the Nissan Leaf weighs over 600 pounds.module for the Nissan Leaf weighs over 600 pounds.

A report from the National Research Council (2010) estimated current battery A report from the National Research Council (2010) estimated current battery costs and projected future costs for plug-in hybrid vehicles. The committee set the costs and projected future costs for plug-in hybrid vehicles. The committee set the most probable current cost for a battery at $875 per kilowatt hour, with $625 per most probable current cost for a battery at $875 per kilowatt hour, with $625 per kilowatt hour being an optimistic estimate. They project battery costs falling by kilowatt hour being an optimistic estimate. They project battery costs falling by 35 percent by 2020 and 45 percent by 2030. At these prices and assuming they scale 35 percent by 2020 and 45 percent by 2030. At these prices and assuming they scale up to the larger battery sizes required for all-electric vehicles, currently the battery up to the larger battery sizes required for all-electric vehicles, currently the battery alone for a mid-sized sedan with a range of 200 miles would cost between $38,000 alone for a mid-sized sedan with a range of 200 miles would cost between $38,000

5 The effi ciency of current electric motors is roughly 80 percentmeaning 80 percent of the energy in electricity goes to moving the vehicle, while current internal combustion engines are in the low 20 percent range. The theoretical bound on effi ciency is roughly 30 percent for the internal combustion engine. For a reasonably accessible explanation, see Johnson (2003) at http://mb-soft.com/public2/engine.html.

106 Journal of Economic Perspectives

and $50,000; the cost of a battery for the mid-sized sport-utility vehicle would be and $50,000; the cost of a battery for the mid-sized sport-utility vehicle would be $53,000$70,000; and a battery for the F-150 would cost between $76,000 and $53,000$70,000; and a battery for the F-150 would cost between $76,000 and $101,000. The optimistic values in 2030 for battery costs alone would be $21,000 for $101,000. The optimistic values in 2030 for battery costs alone would be $21,000 for the sedan, $29,000 for the sport-utility vehicle, and $42,000 for the full-sized pick-up the sedan, $29,000 for the sport-utility vehicle, and $42,000 for the full-sized pick-up truck. The lower cost per mile of electric vehicles would offset these higher upfront truck. The lower cost per mile of electric vehicles would offset these higher upfront costs to some extent. The sedan, for example, at average retail electricity rates would costs to some extent. The sedan, for example, at average retail electricity rates would cost 3 cents per mile, compared to roughly 13 cents per mile at a gasoline price cost 3 cents per mile, compared to roughly 13 cents per mile at a gasoline price of $4/gallon and a fuel economy of 30 MPG. However, these savings in operating of $4/gallon and a fuel economy of 30 MPG. However, these savings in operating costs are unlikely to outweigh the upfront costs at any reasonable discount rate costs are unlikely to outweigh the upfront costs at any reasonable discount rate (Anderson 2009).(Anderson 2009).

While all-electric vehicles may not be cost competitive, vehicles that are partly While all-electric vehicles may not be cost competitive, vehicles that are partly propelled by electricity, such as hybrids or plug-in hybrids, may be. Hybrid and propelled by electricity, such as hybrids or plug-in hybrids, may be. Hybrid and plug-in hybrid vehicles economize on battery costs because they use a higher share plug-in hybrid vehicles economize on battery costs because they use a higher share of the batterys capacity for typical driving patterns. Put another way, if a consumer of the batterys capacity for typical driving patterns. Put another way, if a consumer could size the battery in an all-electric vehicle for each specifi c trip, all-electric could size the battery in an all-electric vehicle for each specifi c trip, all-electric vehicles might be cost competitive at current battery prices. To underline this vehicles might be cost competitive at current battery prices. To underline this point, Anderson (2009) calculates that a plug-in hybrid with a 10-mile range is cost point, Anderson (2009) calculates that a plug-in hybrid with a 10-mile range is cost competitive even at battery costs of nearly $2,000 per kilowatt hour. Similar themes competitive even at battery costs of nearly $2,000 per kilowatt hour. Similar themes are echoed in the more comprehensive analysis of Michalek, Mikhail, Jaramillo, are echoed in the more comprehensive analysis of Michalek, Mikhail, Jaramillo, Samaras, Shiau, and Lave (2011).Samaras, Shiau, and Lave (2011).

The National Research Council (2010) battery cost estimates are somewhat The National Research Council (2010) battery cost estimates are somewhat controversial. The estimates accord well with the published cost estimates for the controversial. The estimates accord well with the published cost estimates for the Nissan Leafs battery of $750 per kilowatt hour (Loveday 2010) and are within the Nissan Leafs battery of $750 per kilowatt hour (Loveday 2010) and are within the range of estimates I have seen for the Chevrolet Volts 16-kilowatt-hour battery pack range of estimates I have seen for the Chevrolet Volts 16-kilowatt-hour battery pack ($500$930) per kilowatt hour (Hall and Schoof 2011; Peterson 2011). However, a ($500$930) per kilowatt hour (Hall and Schoof 2011; Peterson 2011). However, a number of industry trade groups argue that their costs are too high (for example, number of industry trade groups argue that their costs are too high (for example, Electrifi cation Coalition 2009a; CalCars 2010). Better Place, a swappable electric Electrifi cation Coalition 2009a; CalCars 2010). Better Place, a swappable electric vehicle battery company, has stated that they are purchasing batteries at $400 per vehicle battery company, has stated that they are purchasing batteries at $400 per kilowatt hour. Other studies estimate much lower prices under hypothetical situa-kilowatt hour. Other studies estimate much lower prices under hypothetical situa-tions. For example, Nelson, Santini, and Barnes (2009) and Amjad, Neelakrishnan, tions. For example, Nelson, Santini, and Barnes (2009) and Amjad, Neelakrishnan, and Rudramoorthy (2010) simulate battery costs as low as $260 per kilowatt hour and Rudramoorthy (2010) simulate battery costs as low as $260 per kilowatt hour using engineering models of production. These results rely heavily on large scale using engineering models of production. These results rely heavily on large scale economies and an assumption that plants operate 24 hours a day. Under these economies and an assumption that plants operate 24 hours a day. Under these assumptions, costs fall by as much as an order of magnitude when production assumptions, costs fall by as much as an order of magnitude when production increases from 10,000 to 100,000 units per year. Figure 4 plots a number of battery increases from 10,000 to 100,000 units per year. Figure 4 plots a number of battery cost estimates for different points in time, as well as the goal of the United States cost estimates for different points in time, as well as the goal of the United States Advanced Battery Consortium, as summarized in the review article by Cheah and Advanced Battery Consortium, as summarized in the review article by Cheah and Heywood (2010); clearly, the estimates show a large dispersion in all years.Heywood (2010); clearly, the estimates show a large dispersion in all years.

The true cost of batteries, both now and certainly in the future, is unresolved. The true cost of batteries, both now and certainly in the future, is unresolved. But these calculations suggest that some major technological breakthrough may be But these calculations suggest that some major technological breakthrough may be needed for electric vehicles to play a large role in reducing oil consumption: either needed for electric vehicles to play a large role in reducing oil consumption: either a much lower-cost battery, or technological breakthroughs that allow reductions in a much lower-cost battery, or technological breakthroughs that allow reductions in the size and/or weight of vehicles, perhaps through the use of polymer, aluminum, the size and/or weight of vehicles, perhaps through the use of polymer, aluminum, or composite body panels. However, technological breakthroughs reducing size or composite body panels. However, technological breakthroughs reducing size

Christopher R. Knittel 107

and weight could also be applied to internal combustion engines and could thus and weight could also be applied to internal combustion engines and could thus have signifi cant effects on oil use in that waywithout leading to greater use of have signifi cant effects on oil use in that waywithout leading to greater use of electric cars (Knittel 2011). Alternatively, the ranges of electric vehicles could end electric cars (Knittel 2011). Alternatively, the ranges of electric vehicles could end up being much shorter than we are accustomed to hearing about. Indeed, the up being much shorter than we are accustomed to hearing about. Indeed, the battery-powered Nissan Leaf is rated at a range of 73 miles. Air conditioning or battery-powered Nissan Leaf is rated at a range of 73 miles. Air conditioning or heatingbecause heat from the internal-combustion engine can no longer be used heatingbecause heat from the internal-combustion engine can no longer be used to heat the interior of the carsignifi cantly reduces this range. to heat the interior of the carsignifi cantly reduces this range. Car and Drivers s road test for the Nissan Leaf fi nds an average range of 58 miles and discusses the road test for the Nissan Leaf fi nds an average range of 58 miles and discusses the effect of heating (Gluckman 2011).effect of heating (Gluckman 2011).

Hydrogen vehicles also take advantage of the higher effi ciency inherent in elec-Hydrogen vehicles also take advantage of the higher effi ciency inherent in elec-tric motors but generate their own electricity via a fuel cell. Support for hydrogen tric motors but generate their own electricity via a fuel cell. Support for hydrogen vehicles has signifi cantly waned over the past decade, but pursuing the possibility of vehicles has signifi cantly waned over the past decade, but pursuing the possibility of a hydrogen-fueled car remains a stated objective of the U.S. Department of Energy. a hydrogen-fueled car remains a stated objective of the U.S. Department of Energy. Hydrogen vehicles use a fuel cell, which uses a proton exchange membrane to Hydrogen vehicles use a fuel cell, which uses a proton exchange membrane to convert stored hydrogen, and oxygen from the surrounding air, into electricity; the convert stored hydrogen, and oxygen from the surrounding air, into electricity; the by-product of this conversion is water. Fuel cells are cheaper than batteries and by-product of this conversion is water. Fuel cells are cheaper than batteries and refueling could be much faster. (Although supporters of batteries sometimes argue refueling could be much faster. (Although supporters of batteries sometimes argue

Figure 4Battery Cost Estimates from the Literature(as summarized in Cheah and Heywood 2010)

Source: Figure 4 reproduced from Cheah and Heywood (2010), The Cost of Vehicle Electrifi cation: A Literature Review.Notes: Figure 4 plots a number of battery cost estimates for different points in time, as well as the goal of the United States Advanced Battery Consortium, as summarized in the review article Cheah and Heywood (2010). Cost estimates are from Anderman (2010), Air Resources Board (2009), Boston Consulting Group (2010) (BCG), Electrifi cation Coalition (2009b), Frost & Sullivan (2009), National Research Council (2010), Ton et al. (2008) (Sandia), Barnett et al. (2009) (TIAX), and Pesaran, Markel, Tataria, and Howell (2007) (USABC). When a range is given in the original source, Cheah and Heywood plot the average. The USABC number is a goal, not a cost estimate.

Anderman

ARB

BCG

Elec. Coalition

Frost & Sullivan

NRC

Sandia

TIAX

USABC goal

2005 2010 2015 20252020

Bat

tery

pac

k co

st, $

/kW

h

$1,800

$ -

$200

$400

$600

$800

$1,000

$1,200

$1,400

$1,600

108 Journal of Economic Perspectives

that you could refuel quickly via a system of swappable batteries.) At present, that you could refuel quickly via a system of swappable batteries.) At present, however, hydrogen refueling is not simple, with some stations requiring special suits however, hydrogen refueling is not simple, with some stations requiring special suits and apparatus.and apparatus.

Detractors of hydrogen vehicles often point to the fact that they are far less Detractors of hydrogen vehicles often point to the fact that they are far less effi cient than electric vehicles on a well to wheel basis; that is, they take more total effi cient than electric vehicles on a well to wheel basis; that is, they take more total energy to travel one mile, because of the energy needed in making the hydrogen. energy to travel one mile, because of the energy needed in making the hydrogen. However, the more relevant question is the relative cost of the two technologies. However, the more relevant question is the relative cost of the two technologies. That is, if the added energy needed to produce hydrogen were free or low-cost to That is, if the added energy needed to produce hydrogen were free or low-cost to society, then the added ineffi ciency would not matter or would matter less. That is society, then the added ineffi ciency would not matter or would matter less. That is not to say hydrogen vehicles do, in fact, have lower costs. For hydrogen vehicles, the not to say hydrogen vehicles do, in fact, have lower costs. For hydrogen vehicles, the relevant costs are: the cost of the fuel cell, the cost of the high-pressure storage tank, relevant costs are: the cost of the fuel cell, the cost of the high-pressure storage tank, the cost of hydrogen, and infrastructure costs.the cost of hydrogen, and infrastructure costs.

The fi rst main cost element for hydrogen-fueled cars are the fuel cells, which The fi rst main cost element for hydrogen-fueled cars are the fuel cells, which are currently quite expensive. A recent U.S. Department of Energy study ( James, are currently quite expensive. A recent U.S. Department of Energy study ( James, Kalinoski, and Baum 2011) estimates that the cost of fuel cells at the current Kalinoski, and Baum 2011) estimates that the cost of fuel cells at the current fairly low production levels are roughly $230 per kilowatt. To understand what this fairly low production levels are roughly $230 per kilowatt. To understand what this means for costs, the Chevy Volt has a 111-kilowatt electric motor, while the Nissan means for costs, the Chevy Volt has a 111-kilowatt electric motor, while the Nissan Leaf has a 80-kilowatt motor. The Volts motor is equivalent to a 149-horsepower Leaf has a 80-kilowatt motor. The Volts motor is equivalent to a 149-horsepower engine, which is about the amount of horsepower from a four-cylinder gasoline engine, which is about the amount of horsepower from a four-cylinder gasoline engine. Manufacturers appear to install fuel cells equivalent to the size of the engine. Manufacturers appear to install fuel cells equivalent to the size of the motor, so the Volt would require a 111-kilowatt fuel cell at a cost over $25,000. motor, so the Volt would require a 111-kilowatt fuel cell at a cost over $25,000. (There is a prototype Toyota Highlander FCV on loan to the University of (There is a prototype Toyota Highlander FCV on loan to the University of California-Davis that combines a same-sized motor and fuel cell. The Honda FCX California-Davis that combines a same-sized motor and fuel cell. The Honda FCX Clarity does so as well.) The alternative is to hybridize the vehicle by combining Clarity does so as well.) The alternative is to hybridize the vehicle by combining a fuel cell with a rechargeable battery back-up. Of course, electric motor and a fuel cell with a rechargeable battery back-up. Of course, electric motor and fuel cell combinations with horsepower levels comparable to larger vehicles would fuel cell combinations with horsepower levels comparable to larger vehicles would need to be correspondingly much larger.need to be correspondingly much larger.

As with some of the literature on battery costs, a number of papers on the As with some of the literature on battery costs, a number of papers on the future costs of fuel cells are built on assumptions of large scale economies. Using future costs of fuel cells are built on assumptions of large scale economies. Using engineering-economic simulation models, the U.S. Department of Energy study engineering-economic simulation models, the U.S. Department of Energy study assumes a scale economy elasticity of 0.2, and thus simulates that a fuel cell manu-assumes a scale economy elasticity of 0.2, and thus simulates that a fuel cell manu-facturer producing 500,000 units per year could do so at an encouraging cost of facturer producing 500,000 units per year could do so at an encouraging cost of $51 per kilowatt ( James, Kalinoski, and Baum 2011). Given the size of the possible $51 per kilowatt ( James, Kalinoski, and Baum 2011). Given the size of the possible gains from economies of scale and learning-by-doing, more studies along these lines gains from economies of scale and learning-by-doing, more studies along these lines would seem to be an important area for future research.would seem to be an important area for future research.

The second major cost component for a hydrogen vehicle is the storage The second major cost component for a hydrogen vehicle is the storage tank. Hydrogen is ideally stored as a liquid under pressure because this has the tank. Hydrogen is ideally stored as a liquid under pressure because this has the highest energy density. BMW recently demonstrated a hydrogen vehicle with liquid highest energy density. BMW recently demonstrated a hydrogen vehicle with liquid storage. However, storing hydrogen as a liquid faces major obstacles, as the National storage. However, storing hydrogen as a liquid faces major obstacles, as the National Research Council (2004) study points out. For example, the liquid must be kept at Research Council (2004) study points out. For example, the liquid must be kept at 252 degrees Celsius, and the liquid storage tanks currently cost roughly $500 per 252 degrees Celsius, and the liquid storage tanks currently cost roughly $500 per kilowatt hour of energy stored, with the next generation perhaps dropping the kilowatt hour of energy stored, with the next generation perhaps dropping the cost to roughly $100 per kilowatt hour (Brunner 2006). Again using 60 kilowatt cost to roughly $100 per kilowatt hour (Brunner 2006). Again using 60 kilowatt hours as a reasonable guideline for a mid-sized sedan that can travel 200 miles, the hours as a reasonable guideline for a mid-sized sedan that can travel 200 miles, the

Reducing Petroleum Consumption from Transportation 109

storage tank alone would cost $30,000 using current technology and $6,000 using storage tank alone would cost $30,000 using current technology and $6,000 using the projected next-generation technology.the projected next-generation technology.

Thus, absent a major technological breakthrough in liquid storage, hydrogen Thus, absent a major technological breakthrough in liquid storage, hydrogen is likely to be stored as a compressed gas, which either increases the space required is likely to be stored as a compressed gas, which either increases the space required for the storage tank or reduces the range of the vehicle (Ogden et al. 2011). Costs of for the storage tank or reduces the range of the vehicle (Ogden et al. 2011). Costs of compressed storage tanks, if produced at a large scale, might fall between $15 and compressed storage tanks, if produced at a large scale, might fall between $15 and $23 per kilowatt hour of energy (Ogden et al. 2011). Therefore, the storage tank for $23 per kilowatt hour of energy (Ogden et al. 2011). Therefore, the storage tank for a 3,000-pound sedan with a range of 200 miles would cost between $900 and $1,400. a 3,000-pound sedan with a range of 200 miles would cost between $900 and $1,400. However, gas storage tanks face durability issues, which are addressed by making the However, gas storage tanks face durability issues, which are addressed by making the tanks larger. Indeed, the volume of a tank of this size is large enough that manufac-tanks larger. Indeed, the volume of a tank of this size is large enough that manufac-turers are likely to design the vehicle around the tank (National Research Council turers are likely to design the vehicle around the tank (National Research Council 2004). However, if the estimated scale economies truly exist for both fuel cells and 2004). However, if the estimated scale economies truly exist for both fuel cells and storage tanks, the combined cost of the fuel cell and storage tank for a hydrogen storage tanks, the combined cost of the fuel cell and storage tank for a hydrogen vehicle have the potential to be much cheaper than the battery required for an vehicle have the potential to be much cheaper than the battery required for an electric vehicle.electric vehicle.

The third